Inr .I Radmrron 0ncol11~~~ Biol Phw Vol. Printed in the U.S.A All tights reserved.

22. pp.

t .Cil

0360.3016/92 $5.00 Copyright ic‘ 1992 Pergamon Press Ltd.

999-1008

??Hyperthermia Original Contribution

PARAMETERS PREDICTIVE FOR COMPLICATIONS OF TREATMENT WITH COMBINED HYPERTHERMIA AND RADIATION THERAPY DANIEL S. KAPP, PH.D., M.D., RICHARD S. Cox, PH.D., PETER FESSENDEN, PH.D., JOHN L. MEYER, M.D., STAVROS D. PRIONAS, PH.D., ERIC R. LEE, B.S. AND MALCOLM

A. BAGSHAW,

M.D.

Department of Radiation Oncology, Stanford University School of Medicine,

Stanford,

CA 94305

Pretreatment and treatment related factors were reviewed for 996 hyperthermia sessions involving 268 separate treatment fields in 131 patients managed with hyperthermia for biopsy confirmed local-regionally advanced or recurrent malignancies to ascertain parameters associated with the development of complications. A subset of 249 fields were identified in which multipoint or mapped temperature data were available for at least one treatment session per field. A total of 198 fields involved superficially located tumors (I 3 cm from the surface), whereas 51 fields involved more deeply located tumors. Most of these patients had received extensive prior therapy: 77% had surgery, 75% chemotherapy, 65% radiation therapy and 28% hormonal therapy. They were treated with hyperthermia in conjunction with radiation therapy (244 fields) or hyperthermia alone (5 fields). The hyperthermia treatment objectives were to elevate intratumoral temperatures to a minimum of 43.O”C for 45 minutes while maintaining maximum normal tissue temperatures to I 43°C and maximum intratumoral temperatures to 5 50°C. The hyperthermia was given within 30 to 60 minutes following radiation therapy without the administration of additional analgesics. Hyperthermia treatment regimens using radiative electromagnetic, ultrasound, or radiofrequency interstitial techniques were individualized, with 3 to 4 days between hyperthermia treatments and an average of 3.6 treatments (range l-14; standard deviation 2.2) utilized per field. A total of 38 complications in 33 treatment fields were noted; an incidence of 27/198 (13.6%) for fields with superficially located tumors, and 6/51 (11.8%) in fields with more deeply located tumors. Univariate analyses demonstrated statistically significant correlations tween the maximum tumor temperature (p = O.OOOS),average of the maximum tumor temperatures (p = 0.00 Ie6), the average of the % tumor temperatures > 43.5”C (p = 0.0071), and the average number of hyperthermia treatments (p = 0.033), with the development of complications. The average of the maximum measured tumor temuerature for fields without complications was 44.6’C compared with 45.9”C for fields with complications. The comulication rate increased from 7.5% (9/120) in fields that received one or two hyperthermia treatments to 18.6% (24/129) in fields that received greater than two hyperthermia treatments. Multivariate logistic regression analyses revealed the best bivariate model predictive of the development of complications included average of the maximum tumor temperature and the number of treatments per field (p = 0.00012 for the bivariate model). Using the results of the logistic model, isocomplication curves have been generated indicating the risk of the development of complfcations as a function of maximum intratumoral temperatures and number of hyperthermia treatments. Hopefully, these will aid in the design and analysis of future hyperthermia-radiation therapy clinical trials. Hyperthermia,

Combined treatment, Complications,

Thermal parameters, Predictive factors.

INTRODUCIION

A dramatic increase in the clinical utilization of hyperthermia (HT), usually in conjunction with radiation therapy (XRT), has occurred in the past decade with 11% of radiation oncology practices in the United States reportedly performing HT treatments (5). Encouraging initial results have been obtained following combined modality HT-XRT regimens, particularly in superficially located

recurrences from breast, head and neck, and skin cancers (7). While several parameters have been identified which correlate with clinical response and local control following HT-XRT treatment, including minimum and average intratumoral temperatures and radiation dose (8, 21) far less is known in terms of parameters predictive of treatment complications (3, 4, 6, 13). To ascertain pretreatment and treatment related factors associated with the development of complications, records were analyzed for

This work was presented in part at the 30th Annual Meeting of the American Society for Therapeutic Radiology and Oncology, New Orleans, LA, October 1988. Reprint request to: Daniel S. Kapp, Ph.D., M.D. Acknowledgements-The authors wish to acknowledge the

technical support of Allen W. Lohrbach and to thank Sharon Clarke for preparing this manuscript. This work was supported by NC1 contract CM-17480, and NC1 grants CA-40434 and CA-44665. Accepted for publication 3 September 199 1. 999

1000

I. J. Radiation Oncology 0 Biology 0 Physics

patients who participated in an institutional Phase I evaluation of HT equipment (10). METHODS

AND MATERIALS

Patient selection and hyperthermia treatment A total of 13 1 patients were treated with one or more of 2 1 radiative electromagnetic, ultrasound (US), and interstitial radiofrequency (RF) HT applicators as part of a Phase I evaluation trial at Stanford University from September 198 1 to April 1986 and were observed for treatment related complications. The patient characteristics of this group, including their prior treatments, tumor sites, histologies, and thermal treatment parameters have been previously described (10). Patients selected for this trial had measurable tumors, histopathological proof of malignancy, and a life expectancy of at least 3 months. They had advanced or metastatic disease with no chance of cure, or clinically localized, primary or recurrent disease with a low probability of local control by conventional therapies. The patients were informed of the investigational nature of the HT treatment and of the procedures to be performed, and signed an informed consent form which had been approved by the Stanford Medical Committee for the Protection of Human Subjects in Medical Research. The goals of each HT treatment were to obtain a minimum intratumoral temperature of 43.O”C for 45 min while limiting temperatures in normal tissues to I 43.O”C. Microwave and ultrasound applicators were coupled through temperature controlled circulating deionized water, which served as an intervening bolus between the applicator and the surface. The temperature of the water was adjusted to help maintain temperatures recorded on the skin surface to less than 43.O”C. In addition, power usually was reduced to prevent any monitored intratumoral temperatures from exceeding 50°C and any normal tissue temperature from exceeding 43°C. In cases where the tumor extended through the skin surface, attempts were made to heat the involved skin to the same temperature as the remainder of the tumor, and the temperature of the bolus or the power applied were adjusted accordingly. In general, additional analgesics were not used during treatment, and power was reduced to such a level that treatment was tolerated by the patient. Most patients received XRT in conjunction with HT. The HT started within 30 to 60 min following the XRT, and successive HT treatments were separated by at least 3 days. Tumor and normal tissue temperature monitoring was performed during treatment utilizing commercially available systems as previously described (10). When MW or RF devices were utilized, thermistors* or fluoroptic+ temperature sensors were employed. Thermal mapping was

* BSD Medical Corporation, Salt Lake City, UT 84119. t Luxtron, Inc., Mountain View, CA 94043.

Volume 22, Number 5, 1992

performed manually through transcutaneously placed catheter$, with a minimum of one catheter placed for each treatment to encompass tumor and adjacent normal tissue. At least two maps per catheter were obtained during treatment by moving the temperature probes at 0.5 or 1 cm intervals. One map was obtained shortly after steadystate conditions had been achieved and an additional map was usually obtained after at least 30 min of steady-state temperatures towards the end of the 45 min treatment. Additional temperature probes were placed on the skin in the treatment region for continuous temperature monitoring during treatment. When US equipment was utilized, thermocouple arrays were inserted directly into the tissue or probes were placed within thin walled stainless steel needles for monitoring of temperatures. Study design and definitions For the purpose of this analysis, a treatment field was defined as a discrete region that could be treated with a single applicator. A given field could contain one or more discrete tumors or could be diffusely infiltrated with disease. Each field was categorized according to its anatomical site and depth of tumor involvement as designated previously (17). The anatomical sites included the head and neck, thorax, abdomen, pelvis, and extremities. Tumor depths were stratified as superficial (I 3 cm from surface), eccentric (>3 cm from surface but not including the midline) and deep (~3 cm from surface and including the midline). Significant sequelae of a single HT treatment or from an entire HT treatment course were defined as treatment related complications. Patients were evaluated for complications at least once per week during treatment, at 3 weeks following treatment, at monthly intervals for 2 months, and subsequently at 3-month intervals. Complications were characterized according to normal tissue type or tumor, and their management and outcome were noted. Ulcerations that subsequently developed in tumors that had, at the time of onset of HT treatment, extended into the overlying skin were considered tumor ulcerations, whereas those that occurred in initially intact skin and subcutaneous tissues were called normal tissue ulcerations. Infections that developed during treatment possibly secondary to catheter placements for thermometry were scored as being treatment related complications since it was often impossible to define with certainty the actual cause of the infection (e.g., HT treatment or catheter placement). Other toxicities occurring during each HT treatment (acute toxicities) or within 24 hr of the HT treatment and persisting for greater than 24 hr (subacute toxicities) have been previously reported for this patient group and are not included as complications (10). (For example, superficial blisters noted within 24 hr of treat-

* Deseret Medical,

Inc., Sandy, UT 84070.

Complications of hyperthermia and radiation therapy 0 D. S.

ment all resolved spontaneously and were therefore classified as subacute toxicities and not as complications.) Equipment selection and thermal characterization of treatment The selection of applicators to treat a given field was based on tumor site, size, and nature of the surrounding normal tissues. Where possible, one or more treatment devices were chosen for comparison. A detailed description of the MW, US, RF-interstitial and radiative electromagnetic devices employed has been presented previously (10). Individual HT treatments and averages over all HT treatments for a given field were characterized with thermal parameters based on mapped or multipoint intratumoral temperature measurements (10). For example, Tave is defined as the average of all intratumoral temperatures measured during a given HT treatment in a field. Since temperature maps were obtained at least two times along the same catheter during a given 45 min HT treatment, this represents both a spatial and temporal average. The parameter Tave is defined as Tave averaged over all treatments for that field. Similarly, Tmax is defined as the maximum intratumoral temperature measured during all maps or among all multipoint recordings for a given HT treatment, and Tmax as the average of all Tmax for HT treatments in a given field. Analogous definitions apply for minimum intratumoral temperatures (Tmin) and the average of the minimum intratumoral temperatures (Tmin) for a given field. In addition, the maximum Tmax for the entire HT course for a given field was also determined (max Tmax). This represents the highest recorded intratumoral temperature for all HT treatments in a given field. To further characterize the temperature distributions obtained the parameter ‘%T > 435°C’ was defined as the percentage of intratumoral temperatures monitored that were greater than 43.5”C for a given treatment, and %T > 43.5”C the average of this percentage taken over all HT treatments for a given field. The maximum %T > 43.5”C of the entire treatment course for a given field (i.e., the highest %T > 43.5”C for all treatments for a given field) was also determined. Similarly, %T < 41 .O”C and %T < 4 1.O”C were utilized to indicate the percentage of all monitored intratumoral temperatures less than 4 1“C for a treatment, and its calculated average for all treatments for a given field, respectively. For fields in which normal tissue interstitial temperatures were measured, Tmax and Tmax. for normal tissue were calculated in an analogous manner. Patient characteristics and treatment employed The patient characteristics of the entire cohort of I3 1 patients, with 268 separate treatment fields and 996 HT treatments have been previously summarized (10). For the purpose of this more detailed analysis of factors associated with complications, a subset of this cohort was selected including only fields for which multipoint or

1001

KAPP et al.

mapped temperatures were available for at least one HT treatment session per field. A total of 124 patients with 249 fields who underwent 899 HT treatments were identified meeting this additional selection criteria. One hundred and fifty fields contained adenocarcinomas, 37 metastatic melanomas, 29 squamous cell carcinomas, 13 sarcomas, and 20 fields contained tumors of other histologies. The anatomical regions and cla$sification of tumor depth for the 249 fields included: 41 ‘superficial head and neck; 6 deep head and neck; 120 superficial thorax; 2 eccentric thorax; 2 eccentric abdomen; 10 superficial pelvic; 14 eccentric pelvic; 25 deep pelvic; 27 superficial extremities; 1 eccentric extremity; and 1 deep extremity. The relatively high proportion of adenocarcinomas (60.2%) and fields involving the superficial thorax (48.2%) reflects the relatively large number of padients with chest wall recurrences from breast cancer included in this study. Of this cohort of 249 fields, 65% had received prior XRT, 77% had prior surgery, 75% prior chemotherapy (often with multiple treatment regimens)1 28% prior hormonal therapy, 8.0% prior immunotherapy, 4.0% prior radiation sensitizers, and 6% had prior hT treatments. Concurrent XRT was administered to 2b4 (98%) of the fields; HT alone was administered in two superficial fields, and in 3 eccentric or deep fields. Concurrent hormonal therapy was utilized in 9% of the fields, chemotherapy in 3% of the fields, and radiation sensitizers in 1% of the fields. All but 33 of the 249 fields in the principle study cohort had multipoint or mapped temperatures ‘available for all HT treatments per field. This subset of 2 16 fields was separately analyzed for pretreatment and treatment factors associated with complications. The applicators employed for the HT treatments are listed in Table 1. The HT sessions in 190 of the fields (76%) were delivered with one kind of applicator per field, while the remaining 59 fields (24%) received treatment with two or more different applicators during their treatment course.

Statistical analyses The association of various pretreatment and treatment parameters with the development of a complication within a treatment field was investigated by contingency tables or the Student’s t-test. In fields with multipoint or mapped thermometry data for at least one HT treatment, the correlation between pretreatment and treatment parameters with complications was investigated by means of multivariate logistic regression (2). If p is the probability of developing any complication in a treatment field, then the logarithm of the odds p/( 1 - p) is assumed to be a linear function of the covariates, xi, as follows:

log(Pll - P) =

PO +

i

i=l

Pi%.

(1)

1002

1. J. Radiation Oncology 0 Biology 0 Physics

Table 1. Hyperthermia applicators employed in fields in which at least one treatment per field had multiple point or mapped intratumoral temperature measurements Number of fields

Device Waveguide* Single horn* Cylindrical waveguide+ Spiral microstrip S-03* Spiral microstrip S-06* Spiral microstrip S-08* Scanning single spiral# Scanning double spiraP Surface ultrasound transducer U-04* Surface ultrasound transducer U-08* Annular phased array* Isospherical ultrasound device* Interstitial RF system§ Multiple devices used per field Total

3 9 I 1 3 26 23 71

6 1 25 7 14 59 249

* MA 150 Waveguide, MA 201 Single Horn, and Annular Phased Array System, BSD Medical Corporation, Salt Lake City, UT. + Varian Associates, Palo Alto, CA. * See Kapp et al., (10). 5 OximetrixJnc., Mountain View, CA.

The p’s, the parameters of the model, were determined from the patient data by maximizing the logarithm of the likelihood function. Once the p’s were determined, the model was employed to generate isocomplication curves based on the significant prognostic factors. For example, assuming a value for p, xl may be expressed as a function of the other covariates:

x1 = [l”f3(P11 - PI - PO- i: PixillPI.

(2)

i=2

The corresponding variance in xl may be obtained from

Volume 22, Number 5, 1992

thermal measurements and 12.9% for fields with complete temperature data for all treatments. Five fields had two complications in each field; four of those were in superficially located fields, one was in a field with a deep-seated tumor. One hundred and ninety eight fields with multipoint or mapped thermal data for at least one treatment per field were treated with HT for superficially located tumors. A total of 31 complications were noted in 27 of these fields. These included 17 tumor ulcerations, five normal tissue ulcerations, four infections, two fields with induration and fibrosis, one field with persistent edema, one field with persistent pain, and one field with an erythematous skin reaction which developed 18 months after therapy (Table 2). Of the four fields with two complications per field, three had both tumor ulceration and infection, and one had normal tissue ulceration and edema. All 27 superficial treatment fields that developed complications had been treated with HT plus XRT; neither of the two superficial fields treated with HT alone developed complications. Eight of the 17 tumor ulcerations healed spontaneously; seven persisted without additional need for treatment, one patient required medical intervention, and one patient with extensive tumor ulceration of a chest wall recurrence required chest wall resection and reconstruction. Three of the five normal tissue ulcerations healed spontaneously, one required surgical debridement, and one persisted without further treatment. All four local in-the-field infections resolved on antibiotic treatment. The one field that evidenced an erythematous skin reaction at 18 months, developed a tumor recurrence within the field one year later. Seven complications were noted in the 51 fields with multipoint or mapped thermal data treated for eccentrically located or deep seated tumors (Table 3). These included two tumor ulcerations, one infection, one persistent fibrosis, and one field with edema. Both tumor ul-

Table 2. Complications in 198 fields treated with hyperthermia for superficially located tumors

where the Vij are the elements of the variance-covariance matrix obtained in the fitting procedure. RESULTS Complications A total of 38 complications

were noted in 33 fields. All of the complications noted were among the 249 fields for which multipoint or mapped thermal data were available for at least one treatment per field. Twenty-eight fields had complications in the subset of 2 16 fields for which temperature data were available for all treatments. The overall complication rate per field was 12.3% for all 268 fields, 13.3% for the 249 fields with multipoint or mapped

No. complications

Type Tumor ulcerations Normal tissue ulcerations Infections Induration and fibrosis Edema Persistent pain in field Late erythematous skin reaction in treatment field (at 18 months)

17* 5* 4* 2 1 1

Total complications:

31

Total complication

rate per field:

1

27/198 = 13.6%

* Four fields had two complications per field: three had both tumor ulceration and infection, one had normal tissue ulceration and edema.

Complications of hyperthennia and radiation therapy 0 D. S. Table 3. Complications for eccentric

No. complications

Type Tumor

in 5 1 fields treated with hyperthermia or deeply located tumors*

ulcerations

2+

I

Required surgical repair (1) persisted ( 1) Responded to antibiotics Persisted Resolved spontaneously Required surgery

1’

Patient died

Infection

1+

Edema Fibrosis

1 1

Jejunocutaneous

Outcome

fistula

and abdominal wound dehiscence Possible bowel perforation Total complications:

7

Total complication rate per field: 6/5 1 = 11.8% * Located greater than 3 cm from the surface. +One field had both tumor ulceration and infection. * Patient received hyperthermia alone.

ceration and infection developed in one field. One patient developed a jejunocutaneous fistula and abdominal wound dehiscence, and persistent tumor may have contributed to this complication. One patient was noted on X ray studies to have developed free air under the diaphragm which was possibly secondary from a bowel perforation. This patient also had extensive persistent tumor which may have caused this complication. The outcomes

1003

KAPP et al.

of patients with complications are noted in Table 3. Five of the fields with complications were noted among the 48 fields with eccentric or deep-seated tumors who had received HT plus XRT. One complication (possible bowel perforation) was noted among the three fields treated with HT alone. Univariate analyses The association of multiple pretreatment and concurrent treatment factors with the development of complications was examined for the 249 fields in which multipoint or mapped thermal data were available for at least one HT treatment per field and separately for the 2 16 fields with thermometry available for all HT treatments. The average and range of patient pretreatment and thermal parameters for both groups are summarized in Table 4. No statistically significant differences were noted in tumor site, depth (superficial, eccentric, or deep), histology, prior chemotherapy, prior hormonal therapy, prior XRT, prior immunotherapy, prior HT, concurrent hormonal therapy, or concurrent surgery in fields with or without the complications (data not shown). Fields that had undergone previous surgery had a higher likelihood of developing complications than those with no prior surgery (16% vs. 3.5%, respectively, p = 0.025). No correlation was noted between tumor response or local tumor control, and the incidence of complications. A comparison of the 33 fields with one or more comolications to the 2 16 fields without complications revealed a statistically significant higher max Tmax (p = 0.0002), Tmax (p = O.OOOS), %T > 43.5”C (a = 0.0127),

Table 4. Parameters for the fields in which multiple point thermometry or temperature maps were obtained for at least one treatment (249 fields) or for all treatments (216 fields) Fields with thermometry at least one treatment

Parameter Prior radiation dose (Gy)* Karnofsky performance status (%) Number of hyperthermia treatments Concurrent radiation dose (Gy)* Maximum tumor temperature (“C) Maximum recorded tumor temperature (‘C)+ Average tumor temperature (“C) Minimum tumor temperature (“C) % tumor temperature > 435°C Maximum % tumor temperature > 43s”c+ % tumor temperature < 4 1.O”C Maximum normal tissue temperature (“C)*

Average f SD 56.7 78.4

for

Range (minimum-maximum)

f 14.5 ? 14.4

10-105 35-100

Fields with thermometry for all treatments

Average + SD 56.9 79.5

Range (minimum-maximum)

f 14.7 f 13.5

IO-105 35-100

3.61 ? 2.23 33.1 f 14.4 44.8 f 2.1

1-14 5.4-82.0 38.0-53.6

3.13 ?I 1.8 32.7 f 14.1 44.8 + 2.1

l-9 5.4-82.0 38.0-53.6

45.9 42.0 39.4 12.5

f 2.6 + 1.3 f 1.2 f 18.2

38.3-59.5 37.6-47.1 35.9-43.5 O-80.0

45.9 42.1 39.5 11.8

? 2.4 + 1.4 + 1.3 t 17.9

38.3-54.1 37.6-47.1 35.9-43.5 O-80.0

34.9 11.3

-t 25.7 f 17.6

O-100 O-100

43.5 11.0

+ 25.0 + 17.6

O-100 O-100

42.1

+

36.3-47.2

42.1

+

* For fields that received radiation therapy. + Of any treatment for a given field. * For 193 fields that had normal tissue interstitial

2.0

measurements

taken.

2.0

36.3-47.2

1004

I. J. Radiation Oncology 0 Biology 0 Physics

max %T > 43.5”C (p = 0.0158), and average number of HT treatments (p = 0.024) in fields in which complications were noted (Table 5). Nine of 120 fields (7.5%) which received one or two HT treatments developed one or more complications in comparison to 24 of 129 fields ( 18.6%) that received greater than two HT treatments (p = 0.0 17). No significant differences were noted in Tave, Tmin, average prior XRT dose, average concurrent XRT dose, average initial tumor volume, average Karnofsky performance status, %T < 4 1“C, and average maximum normal tissue temperature between fields with or without complications. Very similar results were noted for the average values of the pretreatment and treatment parameters for the subset of 2 16 fields with complete temperature measurements for all fields as for those in the larger cohort of 249 fields (Table 4). Also, the same parameters were noted to be associated with the development of complications when comparisons were made between fields with or without complications in this subgroup (Table 5). Therefore, further logistic analyses were performed only for the cohort of 249 fields. Logistic regression analysis

Pretreatment variables included in the logistic analysis of complications were prior radiation dose, histology, tumor volume, anatomic site, depth of tumor (superficial, eccentric, or deep) and Karnofsky performance status. Treatment parameters included concurrent radiation -dose, max Tmax, Tmax, Tave, E, %T > 43.5”C, max %T > 43.5”C, %T < 41.O”C, and number of HT treatments. Since the seven thermal parameters are correlated,

Volume 22, Number 5, 1992

only models containing one thermal covariate were accepted. Only three models involving a single covariate were found to be significant at p I 0.05: max Tmax (p = O.OOOS),Tmax (p = 0.0006), and number of HT treatments (p = 0.033). Only one model that included two covariates provided a better fit for prediction of complications than max Tmax alone: Tmax and number of HT treatments (p = 0.00012). Additional covariates did not result in a better model so the best model is given by: log(p/l - p) = -18.4

+- 4.5 + (0.348 +- 0.98) (Tmax)

+ (0.202 + 0.079) (No. HT treatments).

(4)

A scatter plot of the values of Tmax and number of HT treatments for each of the 249 fields included in the analysis is shown in Figure 1. Fields without complications are shown by a filled circle (0); those that developed infection by (+); and those with other complications by (X). Also plotted are lines representing the bivariate model predictions for the 5% (lowest line), lo%, 20%, and 40% (uppermost line) probability levels for developing complications for this model. The corresponding error bars shown in Figure 1 represent the square root of VAR (Tmax) according to Equation (3) at the value where the number of heat treatments data points.

is equal to its average over the

DISCUSSION

HT in combination with XRT has become increasingly utilized over the past decade in the management of local-

Table 5. Comparison of fields with one or more complications to those without complications Averages: for all fields with temperature data for at least 1 treatment

Parameter Maximum recorded tumor temp* Maximum tumor temp % tumor temp > 43.5”C Maximum % tumor temp > 43.5T* Number hyperthermia treatments Average tumor temp Prior radiation dose+ Initial tumor volume* Minimal tumor temp Concurrent radiation dose+ Kamofsky performance status % tumor temp < 4 I “C Maximum normal tissue temp

With complications (33 fields)

Averages: for all fields with temperature data for all treatments With complications (28 fields)

Without complications (2 16 fields)

p-value

47.5”C 45.9”C 19.0%

45.7”C 44.6”C 10.7%

0.0002 0.0005 0.013

47.7 46.1”C 21.4%

44.9% 4.42 42.4”C 59.9 Gy 53.0 cc 39.2”C 34.4 Gy 78.8% 12.4% 42.2”C”

33.3% 3.49 41.9T 56.2 Gy 187 cc 39.4”C 32.9 Gy 78.3% 10.7% 42.1 “C”

0.016 0.024 0.083 0.26 0.30 0.45 0.59 0.86 0.61 0.97

47.6% 3.79 42.6”C 60.3 Gy 57.6 cc 39.4”C 33.7 Gy 78.4% 14.5% 42.1 ‘T**

* Of any treatment for a given field. + For fields that received radiation therapy. * For fields with measured tumor volumes. 8 For 193 fields that had normal tissue interstitial measurements taken. ** For 173 fields that had normal tissue interstitial measurements taken.

Without complications ( 188 fields) 45.6 44.6”C 11.2% 34.1% 3.03 42.O”C 56.3 Gy 163.4 cc 39.6”C 32.6 Gy 79.7% 10.9% 42.1 “C**

p-value < 0.0001 0.00 I 0.0050 0.0073 0.038 0.058 0.27 0.42 0.48 0.70 0.65 0.31 0.99

Complications of hyperthermia and radiation therapy 0 D. S. KAPP et al.

PROBABILITY

OF COMPLICATIONS

- without complication + infection x other complication

45

1

f”x;,,T’

. ..I

;

p

!

1

“‘.... ....... 40%

.......

i

Ia

8

1

I

10

0

5

NUMBER

]

10

11

I

10

OF TREATMENTS

Fig. 1. Risk of developing complications as a function of average of maximum intratumoral temperatures (Tmax) and number of hyperthermia treatments for the 249 fields for which multipoint or mapped thermal data were available for at least one hyperthermia treatment per field. Fields without complications are indicated by (0); those with infections by (+), and those with other complications by (X). The four diagonal lines represent the bivariate logistic mode1 predictions of the 5% (lowest line), lo%, 20%, and 40% (uppermost line) probabilities for developing any complication. The confidence bands indicated for each of these four lines represent the square root of VAR (Tmax) at the value where the number of heat treatments is equal to its average over the data points.

1005

regionally advanced or metastatic malignancies, particularly in the treatment of superficial recurrences in previously irradiated fields. While the optimum HT-XRT regimens remain to be defined, numerous retrospective analyses have shown significant correlations between the average of the minimum or average intratumoral temperatures and tumor response (8). Few analyses have attempted to identify parameters predictive of complications of therapy (3, 4, 6, 11, 13). Direct intercomparisons of the available reports are hindered by the limited tumor and normal tissue thermometry employed ( 12); the varying definitions and scoring systems for toxicity and/or complications of treatment; the widely differing applicators used with or without the capability of skin cooling; the variable use of regional or systemic analgesics; the differing formulations characterizing the thermal exposure of the tumor and normal tissues ( 16); the varying radiation and HT treatment parameters; and the diversity of tumors treated - including tumors of different species, sizes, sites, and depths from surface. Table 6 summarizes the thermal parameters associated with “complications” following combined HT-XRT treatment regimens in clinical trials. An initial study of superficially located tumors, heated without surface skin cooling and with limited thermometry, reported that the maximum tumor temperature measured during HT correlated strongly with complications resulting from treatment (13). Acute blistering was included among the complications although these blisters usually healed within 10 days. The frequency of blistering was low (9.1 %I) for maximum tumor temperatures I 42.5”C, and increased (to 53.6%) for maximum tumor temperatures between 42.6”C and 43.9”C. Burns, many of which required up to one month for healing, were not noted at maximum tumor temperatures < 44°C (O/39 tumors), but developed

Table 6. Thermal parameters associated with complications following combined radiation therapy and hyperthermia treatment regimens Factors associated with complications Authors, year Luk et al. ( 13) 1981 Dewhirst and Sim (3) 1984

Number of fields 37 116

Tumor type

Treatment conditions

Human, superficial

No skin cooling

Max tumor temp

Increase in burns if 2 44°C

Dog and cat

General anesthesia

Non-site specific aver max temp (in Eq 43) Aver max skin heat dose per treatment (in Eq 43) Total mean and maximum tumor Eq 43 minutes

Increase in skin infarcts; increase in complications leading to IQSSof life or limb Increase in severe skin reactions

Howard et al. (6) 1987

20

Human,

superficial

Skin cooling

Seegenschmiedt et al. (19) 1988

27

Human, superficial CW recurrences from breast CA

Skin cooling

Kapp et al. present

249

Human,

Skin cooling

all depths

study

Max = maximum; wall; CA = cancer.

Aver = average; Temp = temperature;

Relationship

Factor

Aver max tumor temp, max tumor temp, and number HT treatments

Eq 43 = equivalent

minutes

Increase in acute complications (subcutaneous necrosis and blisters) Increase in complications with increasing max temp and number HT treatments

at 43°C; HT = hyperthermia;

CW = chest

1006

I. J. Radiation Oncology 0 Biology 0 Physics

in 55.6% (5/9) of tumors with maximum tumor temperatures r 44°C. A detailed analysis of complications following combined HT-XRT treatments in 116 spontaneously occurring tumors in dogs and cats revealed a statistically significant correlation with the non-site specific average maximum tumor temperature (expressed in equivalent minutes at 43°C Eq 43) (3). All treatments were administered under general anesthesia. Skin infarcts were noted in 45% of the animals in which intact skin was in the heated field. Their incidence increased as a function of non-site specific average maximum tumor temperature from 19.0% (4/2 1) at I 60 Eq 43 to 80.0% (8/ 10) at > 900 Eq 43. Complications resulting in loss of life or limb also increased with increasing non-site specific maximum tumor temperature with an incidence of 8.2% (5/61) for I 155 Eq 43 compared with 22.4% (1 l/49) for > 155 Eq 43. A further update of this trial confirmed a highly significant relationship between measured intratumoral temperature maximum and the incidence of thermal injury (early normal tissue necrosis, p < 0.0005; persistent normal tissue necrosis, p < 0.0005; and complications leading to extensive reconstructive surgery, loss of eye or limb, or death of the animal, a < 0.0005) (4). The high incidence of thermal injury in this pet trial was believed to be due to either an increased sensitivity of canine skin to thermal damage because of its more segmented blood supply with little collateral circulation, or to the fact that the pets were treated under a general anesthesia eliminating the pain feedback utilized to make power adjustments (4). These investigators also felt that the correlations observed between maximum intratumoral temperatures and thermal injury may be related to unmeasured maxima in adjacent normal tissues. This possibility is supported by the results of a recent clinical trial involving 20 superficial malignant lesions treated with XRT and HT (6). A significant correlation between average maximum skin heat dose per treatment and the total skin reaction score (p < 0.05) was reported. It is noteworthy that in both the pet animal study and in the clinical trial by Howard et al. (6), relatively high dose per fraction radiation treatments (400 or 600 cGy) were utilized in close temporal sequence to the HT treatments. Similarly, high incidences of moist desquamation and fibrosis were noted in patients treated with 500 cGy twice a week and immediate HT in studies reported by Arcangeli et al. (1). Preliminary results from a small clinical trial involving HT-XRT treatment of 27 chest wall fields in patients with extensive superficial recurrences of breast cancer have also revealed a correlation between “acute complications” noted at 1 month and total and mean maximum tumor Eq 43 ( 19). The complications reported included subcutaneous necrosis (11.1%) and superficial blisters (29.6%). For those fields that received a total maximum tumor thermal dose of < 300 Eq 43 or mean maximum tumor thermal dose of < 60 Eq 43, the complications rates at 1 month were 7.1% and 1 l.l%, respectively. Those fields

Volume 22, Number 5, 1992

that received total maximum tumor thermal dose of > 300 Eq 43 or mean maximum tumor thermal dose of > 60 Eq 43 had complications rates of 76.9% and 100% respectively. Interpretation of these results are hampered by the fact that superficial blisters were apparently included among the “acute complications” at 1 month, and that five of the 13 patients in this study received concurrent chemotherapy along with the HT-XRT. Additional treatment related factors have been associated with the development of complications. Lindholm et al. analyzed complications noted among 57 superficially located human tumors treated with HT-XRT (11). Their use of applicators with a cooling water bolus compared to applicators without a water bolus resulted in a decrease in grade 3-4 skin reactions and subcutaneous fat necrosis from 45% and 9% to 13% and 4%, respectively. The use of analgesics during HT treatment was noted to increase both grade 3-4 skin reactions and subcutaneous fat necrosis from a frequency of 29% and 2% (no analgesics) to 44% and 19% (with analgesics), respectively. The high incidence of skin reactions in patients receiving analgesics during treatment is in agreement with the results from pet tumors heated under general anesthesia (3, 4) and in agreement with the significant complications noted in humans treated with interstitial HT where spinal analgesia was often employed (14). A higher incidence of complications was also noted by Lindholm et al. (11) in fields that had received prior XRT (39%) compared to those fields that had not received prior XRT ( 13%). No obvious relationship between total dose of XRT and incidence of complications was noted over the dose range studied (2470 Gy), in agreement with the results presented in our analyses. Not all clinical studies have identified positive correlations between the risk of complications and the pretreatment or thermal parameters. This may reflect, in part, the relatively limited normal tissue and tumor temperature measurements performed (12), the selection of tumors treated, the extent of tumor infiltration into adjacent normal tissues (20) and the intrinsic heat sensitivity of the adjacent normal tissues. In our present study, normal tissue and tumor treatment goals were standardized and skin and normal tissue cooling was utilized wherever clinically indicated. Additional analgesics were not employed during treatment. Measured thermal quantities and their averages were employed to characterize intratumoral temperature distributions rather than secondarily derived quantities, such as equivalent minutes at 43°C ( 16). An attempt was made to better delineate complications of treatment, in distinction to toxicities occurring during and resolving immediately after a given HT treatment (acute toxicities), and to those occurring within 24 hr of the HT session (subacute toxicities) (18). Such definitions are, of course, somewhat arbitrary and will undoubtedly require refinement following additional clinical experience. Furthermore, multivariate logistic regression analyses were performed to

Complications of hyperthermia and radiation therapy 0 D. S.

identify those clinical parameters correlating with complications for the subset of fields in which multiple point or mapped thermal measurements were available (in at least one HT treatment per field). The types of complications noted in the current study appear comparable to those reported previously ( 14, 15). Given the extent of disease in many of the patients treated, the incidence of serious sequelae reported here were felt to be within an acceptable range and somewhat lower than noted in earlier trials ( 14, 15). Initial univariate analysis of pretreatment and treatment factors in our patients revealed that a measure of the maximum intratumoral temperatures, either max Tmax (any treatment for a field) or Tmax, was significantly associated with the risk of complication. This is in agreement with previous studies relating complications to maximum tumor temperatures (3, 4, 13, 19). These results are not surprising since one would anticipate that a complication could result from either a single hot region or repeated heat damage to a region. The average Tmax for fields without complications was 44.6”C compared with 45.9”C in fields with complications. Whether the complications were a direct result of excessive thermal damage to the tumor, however, or were a result of unmeasured high temperatures in adjacent normal tissues (4) is unclear at this time and may be dependent on the exact type of complication considered (e.g., tumor necrosis versus normal tissue necrosis). Monitored normal tissue interstitial temperatures were kept below 43°C whenever possible and there was no difference in maximum normal tissue temperature in fields with complications compared with those fields free of complications (42.2”C vs. 42.1“C, respectively). Similarly, surface temperatures were maintained at I 43°C unless there was tumor involvement of the skin. These findings emphasize the need for more detailed tumor and normal tissue temperature monitoring. In addition, for fields with complications, a higher percentage of measured intratumoral temperatures were greater than 43.5”C (19.0%) as compared to fields without complications ( 10.7%) similarly suggesting the potential risk of excessive heating. Our study also demonstrated a significant correlation of the complication rate with the number of HT treatments. For those fields with multipoint or mapped temperature data, the complication rate increased from 7.5%

KAPP

et al.

1007

in fields that received one or two HT treatments to 18.6% in fields that received more than two HT treatments. Fields without complications received an average of 3.5 treatments compared with 4.4 treatments in fields with complications. These results are consistent with previous trials that have shown a correlation of the total number of 43 Eq minutes for all treatments in a field (19) or total minutes of heat (12) with acute complications or skin reactions, respectively. The independent correlation of both Tmax and number of HT with complication rate was confirmed in our study utilizing multivariate logistic regression analyses. A scatter plot of the values of Tmax and number ofHT treatments for each of the 249 fields, indicating those fields with or without complications (Fig. l), graphically depicts the increasing risk of complications as either or both of these parameters increase. Infectious complications are separately indicated (+) and appear to have occurred at relatively higher Tmax. However, there were too few complications to permit separate statistical analyses of the various subgroups of complications. Employing the results of the best bivariate model, isocomplication lines were plotted representing the model predictions for 5%, lo%, 20%. and 40% probabilities for developing a complication. These curves will aid in predicting the incidence of complications as a function of maximum intratumoral temperatures and number of HT treatments and should assist in the development of useful guidelines for the design of future HT clinical trials. In conclusion, factors associated with the risk of complications were identified in a Phase I HT trial involving a cohort of 124 patients, with 249 HT treatment fields, who underwent a total of 889 HT treatments. Thirty-eight complications were noted in 33 of the fields ( 13.3%). Univariate and multivariate logistic regression analyses have demonstrated a statistically significant correlation between complications and max Tmax, Tmax, and the number of HT treatments, confirming and expanding the results of previous studies. Since HT-XRT trials have revealed an association between tumor response and minimum or average intratumoral temperatures (8, 21) it is suggested that future HT trials should attempt to (a) avoid excessive intratumoral temperatures; (b) obtain high minimal and average tumor temperatures; and (c) identify the minimal number of HT treatments required (9).

REFERENCES 1. Arcangeli, G.; Cividalh, A.: Nervi, C.; Creton, G.; Lovisolo, G.; Mauro, F. Tumor control and therapeutic gain with different schedules of combined radiotherapy and local extemal hyperthermia in human cancer. Int. J. Radiat. Oncol. Biol. Phys. 9: 1125-l 134; 1983. 2. Cox, D. R. Analysis of binary data. London: Chapman & Hall Ltd.; 1970: Chapter 6. 3. Dewhirst, M. W.; Sim, D. A. The utility of thermal dose as a predictor of tumor and normal tissue responses to com-

bined radiation and hyperthermia. Cancer Res. 44(Suppl.): 4772s-4180s; 1984. 4. Dewhirst, M. W.; Sim, D. A. Estimation of therapeutic gain in clinical trials involving hyperthermia and radiotherapy. Int. J. Hyperther. 2: 165-178; 1986. 5. Diamond, J. J.; Hanks, G. E.; Kramer, S. The structure of radiation oncology practices in the continental United States. Int. J. Radiat. Oncol. Biol. Phys. 14: 547-548; 1988. 6. Howard, G. C. W.; Sathiaseelan, V.; Freedman, L.; Bleehen,

1008

7.

8.

9.

10.

11.

12.

1. J. Radiation Oncology 0 Biology 0 Physics N. M. Hyperthennia and radiation in the treatment of superficial malignancy: an analysis of treatment parameters, response and toxicity. Int. J. Hyperther. 3: l-8; 1987. Kapp, D. S. Hyperthermia of superficial malignancies. In: Sugahara, T., Saito, M., eds. Hyperthermic oncology 1988, Vol. 2. Special plenary lectures, plenary lectures and symposium and workshop summaries. London: Taylor & Francis; 1989: 51-56. Kapp, D. S. Areas of need for continued Phase II testing in human patients. In: Paliwal, B. R., Hetzel, F. W., Dewhirst, M. W., eds. Biological, physical and clinical aspects of hyperthermia. Medical physics monograph no. 16. New York: American Institute of Physics, Inc.; 1988: 424-443. Kapp, D. S.; Fessenden, P.; Bagshaw, M. A.; Hahn, G. M.; Samulski, T. V.; Cox, R. S.; Lee, E. R.; Prionas, S. D.; Lohrbath, A. Optimum number of hyperthermia treatments in the combined hyperthermia-radiation therapy treatment of tumors. In: Sugahara, T., Saito, M., eds. Hyperthermic oncology 1988, Vol. 2. Special plenary lectures, plenary lectures and symposium and workshop summaries. London: Taylor & Francis; 1989: 621-622. Kapp, D. S.; Fessenden, P.; Samulski, T. V.; Bagshaw, M. A.; Cox, R. S.; Lee, E. R.; Lohrbach, A. W.; Meyer, J. L.; Prionas, S. D. Stanford University institutional report. Phase I evaluation of equipment for hyperthennia treatment of cancer. Int. J. Hyperther. 4: 75-l 15; 1988. Lindholm, C. E.; Kjellen, E.; Nilsson, P.; Hertzman, S. Microwave-induced hyperthermia and radiotherapy in human superficial tumours: clinical results with a comparative study of combined treatment versus radiotherapy alone. Int. J. Hyperther. 3: 393-411; 1987. Luk, K. H.; Francis, M. E.; Perez, C. A.; Johnson, R. J. Radiation therapy and hyperthermia in the treatment of superficial lesions: preliminary analysis: treatment efficacy, and reactions of skin, tissues subcutaneous. Radiation

Volume 22, Number 5, 1992

13.

14.

15.

16.

17.

18.

19.

20.

21.

Therapy Oncology Group Phase I-II Protocol 78-06. Amer. J. Clin. Oncol. (CCT) 6: 399-406; 1983. Luk, K. H.; Purser, P. R.; Castro, J. R.; Meyler, T. S.; Phillips, T. L. Clinical experiences with local microwave hyperthermia. Int. J. Radiat. Oncol. Biol. Phys. 7: 615-619; 1981. Oleson, J. R.; Sim, D. A.; Manning, M. R. Analysis of prognostic variables in hyperthermia treatment of 16 I patients. Int. J. Radiat. Oncol. Biol. Phys. 10: 223 l-2239; 1988. Perez, C. A.; Emami, B. A review of current clinical experience with irradiation and hyperthermia. Endocuriether. Hyperther. Oncol. 1: 265-277; 1985. Sapareto, S. A.; Dewey, W. C. Thermal dose determination in cancer therapy. Int. J. Radiat. Oncol. Biol. Phys. 10: 787800; 1984. Sapozink, M. D.; Cetas, T.; Cony, P. M.; Eager, M. J.; Fessenden, P. The NC1 Hyperthermia Equipment Contractors’ Group: Introduction to hyperthermia device evaluation. Int. J. Hyperther. 4: 1-15; 1988. Sapozink, M. D.; Gibbs, F. A.; Gibbs, P.; Stewart, J. R. Phase I evaluation of hyperthermia equipment-University of Utah Institutional Report. Int. J. Hyperther. 4: 1I7- 132; 1988. Seegenschmiedt, M. H.; Brady, L. W.; Rossmeissl, G. External microwave hyperthermia combined with radiation therapy for extensive chest wall recurrences. Recent Res. Cancer Res. 107: 147-151; 1988. Valdagni, R.; Amichetti, M.; Pani, G. Radical radiation alone versus radical radiation plus microwave hyperthennia for NS (TNM-UICC) neck nodes: a prospective randomized clinical trial. Int. J. Radiat. Oncol. Biol. Phys. 15: 13-24; 1988. Valdagni, R.; Liu, F. F.; Kapp, D. S. Important prognostic factors influencing outcome of combined radiation and hyperthermia. Int. J. Radiat. Oncol. Biol. Phys. 15: 959-972; 1988.